Watch this video on YouTube
After migrating ETL jobs to run on BigQuery, how can you ensure that the output of the migrated jobs matches the output of the original ones when there is no primary key column available for comparison?